The original plan for the Americans to provide convoy escort on the Canada-Iceland leg of the North Atlantic run starting in April 1941 raised anew the issue of exchanges of intelligence information between the British and Americans. In particular the Americans wanted all available information on German and Italian naval operations in the Atlantic Ocean area. In addition to the growing number of U-boats, which were patrolling ever westward toward Canada, four powerful German warships and five merchant-ship raiders plus twenty-three Italian submarines posed a serious threat to American convoy-escort operations.† Moreover, an even greater menace was in the offing: the super-battleships Bismarck and Tirpitz.
As related, the senior civilian officials of the British and American governments had agreed (in writing, December 1940) to exchange data on breaking Axis codes. What the Americans needed most at that time were the fruits of British successes in breaking German and Italian naval codes. In return, the Americans were to share the fruits of their successes in breaking Japanese codes, in particular the recent sensational cryptographic triumph of duplicating the Purple code machine.
Not all Allied codebreakers on both sides of the Atlantic were keen on the proposed exchange. By nature codebreakers are an obsessively secretive lot, loath to divulge technology to those outside of their respective circles. Each side had misgivings about the trustworthiness of the other, and not without reason. After World War I, the British had indiscreetly and boastfully revealed the secrets of Room 40. Codebreaker Herbert Yardley had indiscreetly and boastfully revealed the secrets of American cryptography. Moreover, each side had been breaking codes of the other for years, an activity justified as spying on a potential “enemy.”
That said, it is also probable that some American codebreakers were quite eager to foster the exchange. Most Americans believed the British Isles were still at great risk, that the Germans might invade in the spring of 1941. They thought it was possible that during the winter Blitz, or the intense preinvasion bombing, or the invasion itself, the British codebreaking establishment could be wiped out or compromised. Others believed that based on the success with the Purple machine codes, the Americans could more effectively exploit existing Enigma technology than the British could, especially by bringing into play appropriate sectors of the vast American electronics and calculating-machine industries.
Whatever the case, the Americans initiated the first step in the Anglo-American codebreaking exchange in early 1941. In Chesapeake Bay on January 25, four Americans—two Army, two Navy*—boarded the new British battleship, King George V, which had just brought the new British Ambassador, Edward F. L. Wood (Lord Halifax), to America. These Americans had with them a Purple machine (or perhaps two machines; sources conflict) and other important materials relating to the decoding of Japanese diplomatic and naval transmissions.
The King George V cautiously crossed the Atlantic during the sortie of Gneisenau and Scharnhorst into those waters. She arrived at Scapa Flow in a snowstorm on February 6. A British brigadier, John H. Tiltman, an expert on nonmachine ciphers who had been designated escort for the Americans, transferred them to a cruiser which took them to the Thames Estuary. From there the deputy director of the Government Code and Cipher School, Edward W. Travis, accompanied the party by automobile to the British codebreaking facility at Bletchley Park. Its commander, Alastair Denniston, welcomed the Americans warmly and courteously.
The American party remained in England about five weeks. Historians of Allied codebreaking still dispute exactly what transpired. Beyond doubt, the Americans turned over the Purple machine (or two of them) and instructed the British in its use. It is what the Americans got in return for this extraordinary gift that is disputed. Some historians say the British disclosed all they knew about Enigma and Italian codes; others say that the British did not—that they were secretive, told little, and kept the Americans at arm’s length. The most astute of these historians, Bradley Smith,†leans to the view that the British withheld too much and thereby shortchanged the Americans. What is not in dispute is the fact that the Americans got no Enigma machine or a clone of one in exchange for the Purple machine.
The Purple machine helped the British enormously. With it they read high-grade Japanese diplomatic traffic for the rest of the war. Of particular benefit to the British (and, of course, the Americans) was the Purple traffic between the Japanese Ambassador in Berlin, Baron Hiroshi Ōshima, and Tokyo. In a previous tour of duty in Berlin, Ōshima had gained the confidence of Hitler, Göring, von Ribbentrop, and other leaders in the Third Reich. The Germans again talked freely of their military plans and weaponry. Ōshima frequently relayed the gist of these talks to Tokyo in Purple code, providing the British—and the Americans—a sort of peephole into Hitler’s mind as well as much specific information on German weaponry.
At this time the British did not have a whole lot of Enigma technology to give the Americans, especially in the naval field. With the assistance of the Turing-Welchman bombas and cribs from various sources, the British were able to read Luftwaffe Red consistently, but even that was a tense daily struggle. As has been revealed by numerous British codebreakers,* little to no progress had been made on breaking naval Enigma. Some did not think naval Enigma could ever be broken without the capture of the daily key settings and other aids.
At the time the Americans arrived in England, the British in fact had afoot an elaborate plan to capture naval Enigma materials. This proposed theft was a high-risk proposition not favored by all concerned. If the Germans became aware of or even got a hint of the theft, they might tighten Enigma security and/or increase its complexity, causing the British to lose—and never regain—Luftwaffe Red. However, the depredations of Hipper, Gneisenau, and Scharnhorst and the merchant-ship raiders, the prospect of a sharp increase in the number of U-boats, the looming threat of Bismarck and Tirpitz, and perhaps pressures from Washington for naval intelligence persuaded the British to go ahead with the theft.
The Germans had thinly occupied the Lofoten Islands off Narvik to exploit the local cod and herring fisheries. Advocates of the theft were persuaded that a strong and well-organized commando raid on the islands might result in the capture of KriegsmarineEnigma and other codes. The operation, Claymore, was staged from Scapa Flow on March 1, 1941. The commando force was substantial: 600 men in two channel steamers, escorted by five big Tribal-class destroyers, which were covered by two cruisers and, more distantly, by stronger elements of the Home Fleet.
Landing in the early hours of March 4, the commandos caught the Germans by complete surprise. To cover the real purpose of the raid, the British blew up the fisheries. In the brisk, brief battle, they captured 213 German prisoners, but disappointingly, they found no Enigmas. The five Tribal destroyers attacked German shipping in the fjords, sinking a 9,800-ton fish factory ship, Hamburg, and several coasters. One German armed trawler, Krebs, bravely fought back. Her skipper, Hans Kupfinger, and thirteen other men were killed in the one-sided fight. Boarding the wrecked vessel, three British officers from the destroyer Somali discovered that Kupfinger had thrown his Enigma machine overboard, but he had died before he could destroy all the Enigma documents, two extra rotors, and some German naval hand ciphers.
The British got a priceless haul from Claymore: Enigma key tables and ring and plugboard settings for February 1941. Utilizing this material, in less than a week (by March 10) the codebreakers at Bletchley Park were able to read the whole of Kriegsmarine home waters code, Heimisch, or as the British called it, Dolphin, for the month of February. That month-old traffic was of no immediate tactical value, but since 95 percent of all Kriegsmarine traffic was encoded in Heimisch (Dolphin), it provided insights into past Kriegsmarine operations and administration and encoding procedures, as well as much specific information on U-boats.
At about this same time the British made an important break into a German naval hand cipher, Werftschlüssel (literally, Dockyard Key) or Werft. This was a “pencil and paper” cipher used by all minor naval vessels and port facilities in German and Norwegian waters. An unglamorous section at Bletchley Park, known as “Cinderella,” had been attacking this hand cipher (and others) cryptanalytically with only slight success. However, as Christopher Morris, one of those in the section, has revealed, from “March 1941” the British read Werft traffic “as a rule currently” for the rest of the war. In all, the Cinderella section decoded 33,000 Werft messages over forty-seven months, an average of about twenty-three signals a day.
Inasmuch as many Kriegsmarine messages had to be transmitted in both Enigma and Werft in order to reach the less important ships and the shore stations, the two systems provided cribs into one another. “Indeed,” Morris wrote, “the ‘cross-ruffing’ between the two was for some time the prize exhibit which Naval Section could display to distinguished visitors, such as Winston Churchill….” Transmissions in Werft from big ships to port authorities or to little ships, such as tugs, or from U-boats in Baltic workup to dockyards or to target ships or tugs, enabled Morris and his fellow codebreakers to pinpoint the locations of the big ships and to identify and track newly commissioned U-boats. When there was a dearth of cribs for. Enigma and Werft, the British resorted to “gardening,” or planting mines in specific zones of known German-swept channels. This invariably provided a spate of warning traffic and/or instructions to minesweepers in Enigma and Werft, which could be “cross-ruffed,” providing a new source of cribs, which the British called “Kisses.”
When the four Americans departed Bletchley Park in late March 1941, they were probably aware of the British Enigma thefts in the Lofotens and the solid break into Werft. They did not return entirely empty-handed. One of the Navy delegates, Robert H. Weeks, has revealed that the British gave them a “paper” Enigma machine (i.e., a detailed drawing) with “all the rotor[s] and umkehrwaltz [reflectors] laid out” and some Enigma keys. According to the American codebreaking historian David Kahn,* the two Army delegates, Sinkov and Rosen, wrote in their official report (of April 1941) that “we were invited to ask questions about anything we saw, no doors were closed to us and copies were furnished of any material which we considered of possible assistance to the United States.”
On the other hand, based on independent interviews with Sinkov, two American codebreaking historians stress the point that the British did not—repeat not—reveal how they actually broke Enigma. In his book,† Thomas Parrish, referring to his interview of Sinkov in February 1984, wrote that on “one or two occasions,” Thomas Parrish, referring to his interview of Sinkov in February 1984, wrote that on “one or two occasions,” the British discussed some “cryptanalytic details” of Enigma with the Americans, but this type of presentation was “essentially an account of some British achievements.” Sinkov said, “It was far from enough to enable us to get into the actual process of producing information.” Based on a review of Sinkov’s papers and an interview with him in October 1990, the aforementioned Bradley Smith wrote that the American party was told only “in a general way” about British decryption processes. “They were not allowed to see a bombe or even told of its existence.” Smith concluded: “The American team therefore had no hard evidence to prove the importance of analytic-machine methods in the British cryptanalytic effort, even though they suspected that some machine-calculation method was being used at Bletchley.”
Among those in Washington who felt the British had betrayed the Americans was the U.S. Navy’s senior codebreaker, Laurence Safford. In view of the Navy’s growing responsibilities—and war risks—in the Atlantic at this time, doubtless Safford was under intense pressure to produce information on German and Italian naval operations in that area. He was bitter toward the British and remained so for the rest of his life, later giving vent to his feelings in several published articles and papers.
As a consequence of this “broken deal,” as Bradley Smith characterized this first cryptography “exchange,” American codebreakers had to face the possibility of breaking Enigma on their own. This added a huge new burden to Safford’s work already in progress, such as solutions to the Japanese naval codes. At that time, the Navy’s entire communications-intelligence organization, including radio-intercept stations in the Atlantic and Pacific areas, had only about 550 staffers, of whom only forty-four were commissioned officers.
Safford therefore sought outside help. Chief among those he leaned upon was a thirty-eight-year-old naval reservist, Howard Theodore Engstrom. Although the codebreaking historians have largely overlooked Engstrom, he was to play the key role in the American attack on naval Enigma, an enterprise of vital importance, which in the rush to give all credit to the Poles and British has likewise been neglected.
Engstrom was the middle of three sons of Scandinavian immigrants—a Swedish father and Norwegian mother. His father, a lifeguard in Plymouth and Manomet, Massachusetts, died when he was ten, leaving his mother, a laundress, to raise the boys, the youngest of whom died at age eleven. A brilliant student, Howard graduated from Plymouth High School in 1918, at age sixteen, and from Northeastern University in 1922, age twenty, with a degree in chemical engineering.
Adept at foreign languages and mathematics, Engstrom chose an academic career. He got a master’s degree in mathematics from the University of Maine (1925) and a doctorate in mathematics from Yale (1929). For the next two years (1930-1931) he was a National and International Research Fellow at the California Institute of Technology and in Gottingen, Germany, where he became fluent in German. From 1932, he was an assistant, then associate professor of mathematics at Yale University. In 1935, he married a Finnish-born woman, Karin Ekblom, who had obtained a bachelor of nursing degree from Yale the year before.
In response to a recruiting drive by the Navy’s Radio Intelligence Organization, which Safford commanded from May 6, 1936, Engstrom accepted a reserve commission as a lieutenant, junior grade. From 1936 to 1940, while on active tours of duty in the summers, he had worked sporadically with Safford. On his stint in the summer of 1940 in Washington, specializing in machine ciphers, he had put in “ten hours a day, seven days a week,” he wrote his older brother, Walder, a bank president in Plymouth.
While still at Yale in the early months of 1941, at Safford’s request Engstrom began some preliminary theoretical work on naval Enigma, it is believed. Several months later, in July 1941, he was called to full-time active duty in Safford’s outfit, with the rank of lieutenant. As will be seen, he rose swiftly to higher rank and responsibilities. His codebreaking achievements, for which both the American and British governments decorated him,* produced results on a par with those of the British mathematicians Alan Turing and Gordon Welchman.